The design of macromolecular binding interactions and complexes, and corresponding alteration of binding
specificity, is a challenging endeavor that remains recalcitrant to computational approaches. This is true both
for the creation of protein-protein complexes (which are driven by a enthalpic changes established primarily by
stereochemical complementarity, balanced against large competing entropic changes) and for the redesign of
protein-DNA complexes (which are heavily dependent upon DNA bending, hydrogen-bonds, electrostatic
contacts, and the presence of solvent and counterions throughout the molecular interface).
Over the past several years we have collaborated with several computational groups to help develop and
validate computational approaches for the design and optimization of protein-protein recognition, protein-DNA
recognition, and protein-small molecule recognition. Those studies have contributed to several new
computational engineering approaches, including hybrid strategies that combine ab initio design of protein
folds and binding sites, the ‘Rotamer Interaction Feld’ (RIF) docking protocol for efficient sampling of protein
sequence and conformation, and novel parametric design approaches to create new tandem repeat proteins.
We propose to continue this work through two specific aims to further develop and improve upon
computational approaches for protein design. As part of this project, we will solve atomic resolution crystal
structures of many selected and designed molecular complexes and provide them to our immediate
collaborators as well as to a public structure prediction project, for computational prediction challenges.
Aim 1. We will design and characterize novel self-associating circular tandem repeat proteins (using both de
novo computational design and using high-throughput selections) and then further design them to undergo
ligand-induced protein-protein association. Beyond the challenge of combining protein scaffold design and
ligand binding design, the motivation for this aim is to determine the structural and mechanistic features of
small molecule ligand-binding, and balance of forces, that facilitate ligand-induced protein-protein association.
Aim 2. We will improve our understanding and ability to design novel protein-DNA recognition specificities and
behaviors. To accomplish this, we will: (1) Systematically select and optimize a series of variants of a model
DNA-binding protein, that display altered binding specificity across two regions of partially overlapping
sequential clusters of basepairs and neighboring protein residues. (2) Determine the high-resolution structures
and binding behavior of each construct. (3) Supervise blinded computational efforts, using multiple
approaches, to predict the same structures. (4) Compare and analyze the results of computational predictions
versus multiple computational prediction strategies to define features influencing predictive accuracy.
For both aims, we will further exploit our crystallographic structures by computationally ‘reverse engineering’
each construct using validated protein structures, to further understand the performance of design approaches.